I had been working for only a few months at my first job and it was the first time I could buy a desktop PC without very tight budget constraints. So I thought I’d look for a more midrange GPU for once. I wasn’t convinced it would be worth it but I said fuck it what’s the point of working and making money if I’m scared to spend it on something I want? So I bought an AMD Radeon 5700XT for ~400€ sometime around Christmas 2019. If you’ve been following PC hardware prices in the COVID era you know I’m extremely happy with my decision.

Law enforcement AI is a terrible idea and it doesn’t matter whether you feed it “false facts” or not. There’s enough bias in law enforcement that the data is essentially always poisoned.

The problem with any excuse you make for Elon is that Elon is too stupid to keep his mouth shut and give the excuse any plausibility. After the nazi salute he went on Twitter to make nazi puns about it. It is certain beyond reasonable doubt that he knows exactly what the salute was. Even if you give him the insane benefit of the doubt that it was really “his heart going out” and accidentally looked like the salute, his having shown he knows what it looks like but never stating he does not actually believe in the ideology or want present himself as an ally to nazis is just as damning.

How is this any less meaningful than any other use case? Is downloading a distro to play video games ok? To shitpost on social media? To watch clickbait videos on youtube? Why is this in particular a bad use of resources?

Maybe in some cases. But I’ve been requested by Google support to provide a video for a very simple and clear issue we were having. We have a contract with them and we personally brought up the issue to a Google employee during a call. There was no concern of AI generated bullshit, but they still wouldn’t respond without a video. So maybe there’s more to this trend than what you’re theorizing.

I find that very unlikely to happen. If AI is accepted as fair use by the legal system, then that means they have a motive to keep copyright as restrictive as possible; it protects their work but allows them to use every one else’s. If you hate copyright law (and you should) AI is probably your enemy, not your ally.

Bold of you to assume this wasn’t always the plan for Pokemon Go. A ton of online services are basically designed from the get go to be mass surveillance machines and the founders know they’re eventually going to be sold as exactly that.

I see. Thanks for sharing. This will be good to know next time I’m looking for a printer.

Microsoft has project Silica where they store data in glass. Being electromagnetic field-proof is one of the stated goals.

but they’re a different kind of hassle

Can you elaborate on this? I thought they would be straight up better to work with and I was thinking of buying one in the future. Is it just about the drying up issue you mentioned or are there other drawbacks?

The DMCA takedown seems to be specifically about Ryujinx’s ability to decode ROMs. Circumventing DRM is in fact illegal according to the DMCA so they appear to have a valid argument. However, in their takedown notice they assume that the decryption keys are obtained illegally. I’m wondering if the DMCA forbids extracting the decryption keys (without distribution) from your own legitimately owned Nintendo hardware for personal backup. If so, then the Ryujinx feature might also be defensible.

This also raises the question of whether an emulator could be made to work on already decrypted media and let you figure out how to do that yourself. Nintendo could argue that its main use is still to play illegally decrypted ROMs but the emulator would have a decent defense imo.

Basically, all encryption multiplies some big prime numbers to get the key

No, not all encryption. First of all there’s two main categories of encryption:

• asymmetrical
• symmetrical

The most widely used algorithms of asymmetrical encryption rely on the prime factorization problem or similar problems that are weak to quantum computers. So these ones will break. Symmetrical encryption will not break. I’m not saying all this to be a pedant; it’s actually significant for the safety of our current communications. Well-designed schemes like TLS and the Signal protocol use a combination of both types because they have complementary strengths and weaknesses. In very broad strokes:

• asymmetrical encryption is used to initiate the communication because it can verify the identity of the other party
• an algorithm that is safe against eavesdropping is used to generate a key for symmetric encryption
• the symmetric key is used to encrypt the payload and it is thrown away after communication is over

This is crucial because it means that even if someone is storing your messages today to decrypt them in the future with a quantum computer they are unlikely to succeed if a sufficiently strong symmetric key is used. They will decrypt the initial messages of the handshake, see the messages used to negotiate the symmetric key, but they won’t be able to derive the key because as we said, it’s safe against eavesdropping.

So a lot of today’s encrypted messages are safe. But in the future a quantum computer will be able to get the private key for the asymmetric encryption and perform a MitM attack or straight-up impersonate another entity. So we have to migrate to post-quantum algorithms before we get to that point.

For storage, only symmetric algorithms are used generally I believe, so that’s already safe as is, assuming as always the choice of a strong algorithm and sufficiently long key.

No, the implied solution is to reevaluate the standard rather than hacking around it. The two humans should communicate that the standard works for neither side and design a better way to do things.

This is really funny to me. If you keep optimizing this process you’ll eventually completely remove the AI parts. Really shows how some of the pains AI claims to solve are self-inflicted. A good UI would have allowed the user to make this transaction in the same time it took to give the AI its initial instructions.

On this topic, here’s another common anti-pattern that I’m waiting for people to realize is insane and do something about it:

• person A needs to convey an idea/proposal
• they write a short but complete technical specification for it
• it doesn’t comply with some arbitrary standard/expectation so they tell an AI to expand the text
• the AI can’t add any real information, it just spreads the same information over more text
• person B receives the text and is annoyed at how verbose it is
• they tell an AI to summarize it
• they get something that basically aims to be the original text, but it’s been passed through an unreliable hallucinating energy-inefficient channel

Based on true stories.

The above is not to say that every AI use case is made up or that the demo in the video isn’t cool. It’s also not a problem exclusive to AI. This is a more general observation that people don’t question the sanity of interfaces enough, even when it costs them a lot of extra work to comply with it.

It’s much more complicated than this. Given that models have been shown to spit out verbatim copies of some training material, it can be argued that the weights do in fact encode the material, just in some obfuscated way. Additionally, it can be argued that the output of the model is a derivative copy of the original work regardless of whether the original work can be “found inside” the model weights, just by the nature of the process. As of now, there is no precedent that I know of on whether this constitutes redistribution of copyrighted material.

How many months should he have waited for an authoritative response?

Well, Marcan should wait as long as feels right to him. As I said previously, I’m pretty sure he was already pissed off about previous R4L issues and he didn’t quit because of this alone. I want to be clear that I’m commenting solely on the expectation of a swifter response from leadership in the original email thread and not on Marcan’s decision to step down, which I can’t be the judge of.

So, I expect people in places of power to take their time when they respond publicly to issues like this, for various reasons. Eg:

• they might try to resolve things in private first (seems to be the case)
• they might want to discuss with their peers to double check their decision making and to take collective action, this is especially true if the CoC committee gets involved
• they might want to chime in when people have calmed down and they expect to be able to have meaningful conversations with them

At the very least, I would have waited to see what happens with the patches if I were in his position. The review process, which kept going in the meantime, essentially sets a timer for a decision to be made. In the end, Hellwig’s objections would either be acknowledged as blocking or they would be ignored. In any case there would have been a clear stance from the project’s leadership. It makes sense to me to wait for this inevitable outcome before making a committal decision such as stepping down.

Cristoph Hellwig’s initial message was on 2025-01-08. Marcan’s stepping down was on 2025-02-07. So no, it’s not several months; it’s barely one month. Getting in fights in mailing lists and making social media posts is not everyone’s first reaction and it is arguably not the best reaction, especially for people in places of power. It is silly for Marcan to demand everybody’s reaction to be as loud and as quick as his own.

It was very clear that the reaction was going to be no reaction.

Well, it turns out that the reaction was pretty clear not “no reaction”. That’s the reason this thread we’re talking in exists. Marcan was objectively wrong if he assumed Hellwig’s comments and nack would be accepted. Instead, Hellwig was explicitly called out for having no say on the matter and for producing “garbage arguments”.

Marcan is not the submitter. Unless I’ve missed something, the submitter is still working on the patch.

Marcan was probably fed up and was looking for a reason leave. If that’s not the case, then it’s silly for him to just quit mid-discussion, before it’s even become apparent what the reaction to Cristoph Hellwig’s behavior would be and whether his reply would even be taken into account during the review process.

Arch doesn’t require you to “read through all changelogs”. It only requires that you check the news. News posts are rare, their text is short, and not all news posts are about you needing to do something to upgrade the system. Additionally, pacman wrappers like paru check the news automatically and print them to the terminal before upgrading the system. So it’s not like you have to even remember it and open a browser to do it.

Arch is entirely about “move fast and break stuff”.

No, it’s not. None of the things that make Arch hard for newbies have to do anything with the bleeding edge aspect of Arch. Arch does not assume your use case and will leave it up to you to do stuff like edit the default configuration and enable a service. In case of errors or potential breakage you get an error or a warning and you deal with it as you see fit. These design choices have nothing to do with “moving fast”. It’s all about simplicity and a diy approach to setting up a system.